Objectives: This study aimed to create and analyze the performance of an automated triage tool alerting triage nursing staff and physicians to an abnormal heart rate consistent with septic shock in a pediatric emergency department.
Methods: A computerized best-practice alert (BPA) triage system corrected heart rate for temperature (5 beats per minute for each 1°F above 100°F or 9.6-10 beats per minute for each 1°C > 36°C) and alarmed on tachycardia. If patients appeared ill and/or had medical comorbidities predisposing them to sepsis, a "shock protocol" was activated. Sensitivity was calculated for patients clinically diagnosed with shock during the study period.
Results: During the study period (February to August 2010), the BPA was triggered in 4552 (11.5%) of 39,697 visits. Mean age was 5.4 years (range, 18 days to 18 years); 53% were female. The tool was 81% sensitive in identifying the 210 patients with shock. Missed patients were more likely to be previously healthy (odds ratio, 2.7; 95% confidence interval, 1.2-6.2), younger (5.7 vs 8.7 years, P = 0.004), and less likely to have a malignancy (odds ratio, 0.38; 95% confidence interval, 0.2-0.8). The tool was 89% specific; positive and negative predictive values were 4% and 99.9%, respectively.
Conclusions: The BPA-automated sensitive triage tool, based solely on initial temperature and heart rate, led to the identification of most children with septic shock, even before clinical acumen and laboratory values were incorporated into the diagnostic algorithm.